-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
Sony presenting "Designing a game audio engine for HSA" at AMD's Dev Summit
- Thread starter artist
- Start date
Based on the overview of the session, it sounds like TrueAudio ..I'm guessing that amd has no problems with having sony/ms engineers optimising solutions for their tech
jeff_rigby
Banned
My understanding is that HSA_IL code is not supposed to involve Graphics or Audio. HSA_IL can support cross platform code for Accelerators, GPU as CPU and CPUs.
So I don't know how to take this..... Just portions of the code HSA_IL, HSA efficiencies being used or HSA_IL extensions to support audio?
So I don't know how to take this..... Just portions of the code HSA_IL, HSA efficiencies being used or HSA_IL extensions to support audio?
phosphor112
Banned
Interesting. I want to see how this'll work out since we know the DSP in the PS4 isn't nearly as complex as the X1's.
X1's DSP is mostly dedicated to Kinect processing.
XtremeXpider
Member
Not SHAPE like chip on PS4 confirmed?
Just look in B3D, you are wrong. There are DSPs for Kinect AND there is a chip called SHAPE, not Kinect related.
Just look in B3D, you are wrong. There are DSPs for Kinect AND there is a chip called SHAPE, not Kinect related.
phosphor112
Banned
I understand, people need to keep in mind Sony is going to include voice commands as well. It WILL use some resources if it they want to keep performance comparable.
Only if voice commands are available during non-voice games.
No, it implies this is actually a possibility. (though I'm not holding my breath)
phosphor112
Banned
Well... yeah. Lol, that's the point of my comment.
XtremeXpider
Member
I don't understand how is it a possibility reading the OP. IMO it looks like devs can use the GPU for sound processing.
This is likely to include a ray casting audio engine. Really has nothing to do with xbone chips. Shape chip just mixes/decode audio.
Well, not really, because these voice commands will be used to navigate the settings/menu in the PS4's XMB (or whatever it is called now). Voice commands won't be there during the gameplay, as it seems. (and that's a good thing)
dragonelite
Member
Having access to collision mesh data should help with 3D audio.
Can Crusher
Banned
HSA confirmed!
You are probably confusing HSA with hUMA ..
Slightly off-topic, but have you heard if the API (AstoundSound I believe) is open or did AMD get exclusivity towards it? I'd like to see some proper sound card manufacturers use it.
Well unless I'm mistaken we know both UMA and HSA so doesn't that automatically leave us with hUMA?
Looks like we can definitely exclude a SHAPE-like architecture, at least in its complexity like the X1 but the big question for me is how can the PS4 substitute this with it's own resources? A single CPU Core? Even two? CUs? Sure we don't need the complex Kinect Audio stuff but I'd strongly appreciate if we finally move from flat boring prebaked audio and onto more dynamic, procedural techniques. Didn't Cerny speak of raycasting?
Oh, and the most important thing. We need more complex approaches on HRTF and leave stuff like Dolby Headphone and Co. behind and back to the roots with game engines giving exact coordinates to the sound renderer.
To me HRTF is the future, not only because of VR like Occulus but more because of it's accessibility in contrast to a fully prepared and "modded" room full of 6-8 speakers.
Oh, and the most important thing. We need more complex approaches on HRTF and leave stuff like Dolby Headphone and Co. behind and back to the roots with game engines giving exact coordinates to the sound renderer.
To me HRTF is the future, not only because of VR like Occulus but more because of it's accessibility in contrast to a fully prepared and "modded" room full of 6-8 speakers.
AFAIK no. Didnt NV have their own sound solution at one point? They might resurrect that now.
Always nice to get that confirmation.
Yep, seems so ^^
Huma is part of hsa amd design.
Bandwidth of audio is pretty small compared to video. I guess any GPU should be well suited to HRTF processing, as it's just convolution filtering. Get HRTF interpolation for free as well.
More traditional DSP tasks can be done with a little CPU power. Simple IIR filters and the like. You can run hundreds of filters with a single core.
The DSP in XB1 is very nice but I think PS4 CPU+GPU can cover it pretty well.